Admin Companion 5.0 is here: faster, smarter, and more transparent

: 09 October 2025

We’ve just launched Admin Companion v5.0 — and we’re genuinely thrilled. In our own usage, the improvements have been remarkable. This release brings expert knowledge you can trust, quicker reasoning, and clearer insight into how answers are produced.

🧠 Expert Knowledge + Semantic Search (RAG-powered)

We’ve bundled a curated 100 MB Linux expert corpus to ground responses in professional-grade knowledge. Queries are matched with semantic search (meaning-aware retrieval), and the relevant context is passed to the model to facilitate Retrieval-Augmented Generation (RAG).

Result: answers that are highly accurate.

Why it matters: fewer hallucinations, better citations/context, and stronger recommendations for real-world sysadmin tasks.

📝 Multi-Line Request Editor

Compose and refine complex prompts with full cursor navigation, just like an editor. Submit with Alt + Enter whenever you’re ready.

Why it matters: iterate without losing flow — perfect for long commands, step-by-step runbooks, and incident notes.

⚡ Reasoning Efficiency

We optimized reasoning to reduce token use in specific scenarios — without changing pricing.

Why it matters: faster answers and more mileage from your plan, particularly on longer chains of thought.

💡 Real-Time Command Streaming

Commands and scripts stream as they’re generated, so you can start reviewing immediately.

Why it matters: no more waiting for full output — grab the pieces you need as they appear.

📌 Persistent Background Info

Store reusable system context — installed services, non-standard paths, environment quirks — and have Admin Companion factor it in automatically.

Why it matters: fewer reminders and clarifications; replies that fit your environment.

⏳ Live Reasoning Feedback

Watch the reasoning phase unfold during “Thinking…,” especially helpful with advanced models like GPT-5.

Why it matters: transparency — you’ll see how the model is approaching your task.

FAQ

Does RAG replace the model’s knowledge?
No — RAG supplements it with curated Linux expertise and surfaces the most relevant snippets for your query.
Do I have to enable anything to get streaming?
No. Command/script streaming is built in to v5.0.
Will my plan change?
No. Pricing remains the same.
Can I turn off usage of the curated Linux knowledge?
Yes — use the command line parameters to adjust the usage of the corpus.